A Physics-Based Prediction Engine for the Combustion Predictive Thermometer

The challenge

Combustion, Inc. makes the Combustion Predictive Thermometer (CPT), a wireless thermometer used by home cooks and chefs. The company wanted a physics-based prediction engine that simulates how heat moves through food and uses that to predict cooking times and temperatures. Simulations like this usually run on servers. Combustion wanted it to run on the thermometer itself, on a Cortex-M4 processor with limited memory, computing power and battery life.

What we did

We modelled the physics of heat transfer through food and developed algorithms that predict cooking times and temperatures from it. We then adapted those algorithms to the Cortex-M4, combining algorithm optimization with careful embedded firmware engineering so the heat-transfer simulation runs in real time within the device's limits. Our advanced algorithm work on this project sits at the meeting point of physics and constrained hardware.

The result

The prediction engine runs in real time on the thermometer. It became the foundation for later product updates, including more accurate cooking-time predictions and adjustments for different cooking conditions.

Work With Us